I'm working on a new release of Acovea that's broader in scope (it can optimize things other than compilers, and handles makefiles now, among other things.)
Once I get 5.0 out the door, I'm going to start in on a version 6.0 that collects statistics in a database so you can combine the results from multiple tests across different applications.
Even at commercial companies, QA isn't a "sexy" task. People would rather bang out code than write testing harnesses and run benchmarks.
Also, free software is driven by programmers, who tend to hate QA. Like any artist or craftsman, a programmer hates having their work critiqued. They spent hundreds (or thousands) of hours on a program, only to have someone nit-pick the details and point out the flaws. But for art, "quality" is a subjective quality -- and with software, quality and reliability are tangible quantities that can be measured.
My Acovea project demonstrated the problem. Users of GCC love Acovea; many developers of GCC, on the other hand, seem to treat it is an annoying distraction. Acovea identified more than a dozen errors (some quite serious) in the GCC 3.4/4.0 compilers -- and yes, I did report them to bugzilla. Only a couple of GCC's maintainers have said "thanks."
Not that the cool reception deters me. I have a new version of Acovea in the wings, and will be unleashing it on GCC 4.x Real Soon Now.;)
As a consultant, I've been paid to perform QA work on commercial software packages -- but only one company, and a big one at that, has ever contracted me to QA a free software project.
Right now, free software is about many things, but quality is not job 1. And that needs to change.
I found a copy of Linux Acrobat 7.0 a couple of weeks ago, and using it has been a joy as compared to KPDF, Ghostview, and XPDF. Bookmarks and thumbnails work properly, the layouts look very clean (letter spacing is correct, etc.) -- I am so thankful Adobe finally relased the Reader for Linux.
Now if I could just get the complete Adobe Acrobat 7.0 on Linux, I'd be positively ecstatic.
...it changes it's attitude. An excerpt from my blog entry earlier this week:
When I buy groceries, the digital display at the register is running a Windows application to show my purchases.
At the autoparts store, the inventory system has a Windows interface to a SQL Server database.
The woman who sells me insurance runs a Microsoft Access application on her laptop.
To withdraw money from my bank's ATM machine, I use a Windows interface -- I know, because when the ATM crashes, it displays Windows XP dialog boxes.
I could come up with more examples, but the point should be clear: Businesses survive on vertical market software, and that means Windows. Companies don't buy their core applications at the store -- they go to a value-added reseller (VAR) or independent software vendor (VAR), or even develop the software in-house.
Linux could compete for industry-specific applications, if it had the tools to compete with Access, Visual Basic, and the other facilities that ISVs and MIS departments rely upon. Ease the transition, provide better tools... in some ways, I think that is what Mono was meant to do, by bringing the.NET "platform" to Linux.
If Linux is to become universal (even even significantly competitive) in the business world, it needs to understand how that business software is created.
I'm not a big Le Guin fan, and I looked at The Legend of Earthsea simply as a diversion.
The mini-series was not awful, but it certainly wasn't very good, either. The actors were so understated as to be boring; the only reason I cared about Tinar is because she was cute.;) As for the main character, he was a stereotypical pretty boy; his sidekick Vetch was the traditional pudgy geek. The best character was a dragon, who figures in about three minutes of screen time.
Le Guin should be upset, but not surprised. Publishers, TV execs, and movie makers have always twisted ideas to their own ends; even examples such as Jackson's LOTR do not prevent "the powers that be" from dumbing down artistic vision for mass audiences.
So why do creative people let their worlds be perverted by publishers and movie makers? Because you can't make money if your work doesn't get printed and sold. It's a myth that people will pay artists through online contributions; it just doesn't happen.
I never have mod points when something *good* needs to be modded up.
"igb" is correct; in fact, some cathedrals have never been finished, even though they are quite useful and beautiful! Antonio Gaudi'sLa Sagrada Familia Cathedral in Barcelona is perhaps the perfect example of a fantastic structure that is taking centuries to construct!
ESR should really spend some time understanding the foundations of his metaphors before building his arguments.
Microsoft is afraid -- their revenues are flat or falling in certain markets, the trend toward free software is undermining the value of commercial software, and people are moving to more secure products in light of Microsoft's deficiencies.
Fear and greed are powerful motivating factors, especially when combined. Microsoft can not compete on reliability or uptime or prices, so it can only fight back in one way: by sharing it's fear.
In the real world diverse choices at one level (KDE vs. Gnome, let's say) result in reduced choice at another (I chose KDE but now need to run Eclipse etc.).
Odd -- I've spent an awful lot of time running eclispe on KDE without trouble; perhaps I've been hallucinating?;)
I run plenty of GTK apps (Gnumeric, GIMP) all the time under KDE, and they work just fine. Monocultures are dead ends; moderate diversity is the fuel of innovation.
Otherwise it is too expensive for small / midsize shops to buy the licenses need to ship their QT projects.
Very true.
While I prefer KDE as a working environment, the QT license has left me to develop commercial applications using GTK, simply because the GPL does not impose any financial overhead.
Licensing is a big area of uncertainty, with some projects involving dozens of different "open licenses". Business hates uncertainty, which keeps Microsoft on top. Your average CEO or CFO is concerned with simplicity (one vendor and license) and stability (Microsoft is consistent, if not perfect); even if free/open software is "better" from a technical standpoint, it isn't going appeal to the folks who run corporations.
In the long term, I see KDE being the preferred platform for the free/open/geek crowd, Gnome settling in with the "desktop business" distros like SuSE, and Windows holding its dominance in the corporate world.
I'll never understand the religious wars about these issues. It's technology, folks -- use whatever works for you.
Freedom is predicated on the availability of diverse choice; we need different philosophies and approaches.
For day-to-day work, I use KDE, though I prefer Thunderbird to KMail (or Evolution, which is overkill for my purposes). I've run Gnome quite a bit, too; my Opteron system has both Gnome and KDE installed, and I spend about 90% of my time in the latter. I can live with either one, though I prefer the customization available in KDE.
Gnome and KDE both have high overhead (disk space and processor use) as compared to XFCE, which is the GUI for my dusl 600MHz Pentium 3 and 300MHz Sun Ultra 10.
My Pentium 4 box dual-boots between Gentoo/KDE and Windows XP. I find XP limited in many (many) respects, but some things (games) just work better under Windows.
Who said there are only 2 compilers? He only tested 2, that's all. Here's a bit of a list - and notice that some of these are targeted specifically to scientific computing:
In the article, I explicitly state that I'm open to testing other compilers; in fact, Comeau has already contacted me. I have attempted to obtain review compilers for PGI and Pathscale, without success.
Cheesy passive-agressive self deprecation comedy mixed with a "slip it to 'em that I probably had sex at least once" tangent? Thank god I didn't read your article.
Choice is no substitute for software freedom, in fact they speak to different aims entirely and in computer software choice is not as important as software freedom.
You can not have freedom without choice. Competition in a diverse environment is the engine of evolution; any monoculture stagnates. I no more want a world run by RMS than I do a society that is beholding to Mr. Gates. A pox on both their monopolistic houses.
I believe in the competition of ideas, and strongly support and defend the concept of intellectual freedom. I've debated these issues with RMS directly, so I've been through this war before. Much as I respect his ideals, his implementation is lacking and his approach dogmatic and absolutist. Freedom involves the right to disagree with RMS and his acolytes.
Many argue that we cannot have democracy if we focus on choice: two dominant US political parties have and will cooperate and pursue the same ends (through scheduling exclusionary debates, supporting bills the public doesn't agree with, raising ballot-access levels, making elections prohibitively expensive, etc.)
The current U.S. political system does not offer significant choice, which is precisely why we need diversity. The difference between Demicans and Republocrats is so small as to be meaningless; trading Kerry for Bush is nothing more than changing dance partners while the same tune plays.
What is exactly the point of comparing gcc and g++?
Two reasons:
I'm tracking the recent evolution of GCC, looking for answers to questions like: Are new versions of GCC better than their predecessors?
Different Linux systems use different versions of GCC. My Opteron's Gentoo was compiled with 3.4.1, while my Pentium 4 uses 3.3, for example. Some people stick to older versions for stability or code quality purposes, too.
Most of the compilers you list are not serious contenders. OpenWatcom is planning, but does not yet distribute, a Linux compiler. I found no mention of Linux on the lcc web site. I don't have an Alpha anymore, so the Compaq compiler is not pertinent. I don't have a copy of the Absoft compiler, even though I requested one for review. An interpreter is, as you note, not a compiler. And so on...
Wolfram has some interesting points, but I think his ego has gotten in the way of good science. With his "New Kind of Science", Wolfram hasn't really "invented" anything; he's mostly implying meaning where it may or may not exist.
On the other hand, I do believe that higher order derives from basic, simple, underlying processes that combine in great complexity. Turning that supposition into real science is something I hope to see happen in the next century.
Mr. Ladd, you have just become my official hero. I used to suppose that I was the only socially hermetic bastard who sat around with a compiler-driving keyboard in one hand and an icing-laden spoon in the other; now I know that I'm not alone in the universe.
Well gosh darn, your testamony makes me feel all warm and fuzzy inside. Or maybe that sensation comes from the rum and frosting...
Have you considered including some non-scientific code in your benchmarks, though? There are many of us who write primarily integer and/or string-based code, and don't deal with floating point to any significant extent.
As you noted, the problem is finding such a benchmark that doesn't bring I/O and other external latencies into play. I'm open to suggestions like yours.
I've had problems using Intel to integrate existing libraries. As of version 8.1, the binary compatibility is supposed to be better, but I haven't had a chance to apply it to a really big project. In all fairness, GCC has had some similar problems with the C++ ABI at times.
I'm working on a Fortran 95 comparison, which will include several products.
Most people doing serious numerical work are creating proprietary or in-house applications; few people have the clusters required to run the type of code written in Fortran. So there is a commercial market for Fortran; places like National Laboratories and universities have the money to buy software.
Slashdot Mirror
I'm working on a new release of Acovea that's broader in scope (it can optimize things other than compilers, and handles makefiles now, among other things.)
Once I get 5.0 out the door, I'm going to start in on a version 6.0 that collects statistics in a database so you can combine the results from multiple tests across different applications.
Morton is correct.
Even at commercial companies, QA isn't a "sexy" task. People would rather bang out code than write testing harnesses and run benchmarks.
Also, free software is driven by programmers, who tend to hate QA. Like any artist or craftsman, a programmer hates having their work critiqued. They spent hundreds (or thousands) of hours on a program, only to have someone nit-pick the details and point out the flaws. But for art, "quality" is a subjective quality -- and with software, quality and reliability are tangible quantities that can be measured.
My Acovea project demonstrated the problem. Users of GCC love Acovea; many developers of GCC, on the other hand, seem to treat it is an annoying distraction. Acovea identified more than a dozen errors (some quite serious) in the GCC 3.4/4.0 compilers -- and yes, I did report them to bugzilla. Only a couple of GCC's maintainers have said "thanks."
Not that the cool reception deters me. I have a new version of Acovea in the wings, and will be unleashing it on GCC 4.x Real Soon Now. ;)
As a consultant, I've been paid to perform QA work on commercial software packages -- but only one company, and a big one at that, has ever contracted me to QA a free software project.
Right now, free software is about many things, but quality is not job 1. And that needs to change.
I found a copy of Linux Acrobat 7.0 a couple of weeks ago, and using it has been a joy as compared to KPDF, Ghostview, and XPDF. Bookmarks and thumbnails work properly, the layouts look very clean (letter spacing is correct, etc.) -- I am so thankful Adobe finally relased the Reader for Linux.
Now if I could just get the complete Adobe Acrobat 7.0 on Linux, I'd be positively ecstatic.
Linux could compete for industry-specific applications, if it had the tools to compete with Access, Visual Basic, and the other facilities that ISVs and MIS departments rely upon. Ease the transition, provide better tools... in some ways, I think that is what Mono was meant to do, by bringing the .NET "platform" to Linux.
If Linux is to become universal (even even significantly competitive) in the business world, it needs to understand how that business software is created.
I'm not a big Le Guin fan, and I looked at The Legend of Earthsea simply as a diversion.
The mini-series was not awful, but it certainly wasn't very good, either. The actors were so understated as to be boring; the only reason I cared about Tinar is because she was cute. ;) As for the main character, he was a stereotypical pretty boy; his sidekick Vetch was the traditional pudgy geek. The best character was a dragon, who figures in about three minutes of screen time.
Le Guin should be upset, but not surprised. Publishers, TV execs, and movie makers have always twisted ideas to their own ends; even examples such as Jackson's LOTR do not prevent "the powers that be" from dumbing down artistic vision for mass audiences.
So why do creative people let their worlds be perverted by publishers and movie makers? Because you can't make money if your work doesn't get printed and sold. It's a myth that people will pay artists through online contributions; it just doesn't happen.
I ask this question in all seriousness: What specific benefits would I gain from switching a Linux system to one of the BSDs?
I never have mod points when something *good* needs to be modded up.
"igb" is correct; in fact, some cathedrals have never been finished, even though they are quite useful and beautiful! Antonio Gaudi's La Sagrada Familia Cathedral in Barcelona is perhaps the perfect example of a fantastic structure that is taking centuries to construct!
ESR should really spend some time understanding the foundations of his metaphors before building his arguments.
Microsoft is afraid -- their revenues are flat or falling in certain markets, the trend toward free software is undermining the value of commercial software, and people are moving to more secure products in light of Microsoft's deficiencies.
Fear and greed are powerful motivating factors, especially when combined. Microsoft can not compete on reliability or uptime or prices, so it can only fight back in one way: by sharing it's fear.
I'll let Merkey speak for himself, and let other judge who and what he is...
Odd -- I've spent an awful lot of time running eclispe on KDE without trouble; perhaps I've been hallucinating? ;)
I run plenty of GTK apps (Gnumeric, GIMP) all the time under KDE, and they work just fine. Monocultures are dead ends; moderate diversity is the fuel of innovation.
Very true.
While I prefer KDE as a working environment, the QT license has left me to develop commercial applications using GTK, simply because the GPL does not impose any financial overhead.
Licensing is a big area of uncertainty, with some projects involving dozens of different "open licenses". Business hates uncertainty, which keeps Microsoft on top. Your average CEO or CFO is concerned with simplicity (one vendor and license) and stability (Microsoft is consistent, if not perfect); even if free/open software is "better" from a technical standpoint, it isn't going appeal to the folks who run corporations.
In the long term, I see KDE being the preferred platform for the free/open/geek crowd, Gnome settling in with the "desktop business" distros like SuSE, and Windows holding its dominance in the corporate world.
I'll never understand the religious wars about these issues. It's technology, folks -- use whatever works for you.
Freedom is predicated on the availability of diverse choice; we need different philosophies and approaches.
For day-to-day work, I use KDE, though I prefer Thunderbird to KMail (or Evolution, which is overkill for my purposes). I've run Gnome quite a bit, too; my Opteron system has both Gnome and KDE installed, and I spend about 90% of my time in the latter. I can live with either one, though I prefer the customization available in KDE.
Gnome and KDE both have high overhead (disk space and processor use) as compared to XFCE, which is the GUI for my dusl 600MHz Pentium 3 and 300MHz Sun Ultra 10.
My Pentium 4 box dual-boots between Gentoo/KDE and Windows XP. I find XP limited in many (many) respects, but some things (games) just work better under Windows.
Competition is a good thing.
In the article, I explicitly state that I'm open to testing other compilers; in fact, Comeau has already contacted me. I have attempted to obtain review compilers for PGI and Pathscale, without success.
Sounds like jealousy to me... ;)
You can not have freedom without choice. Competition in a diverse environment is the engine of evolution; any monoculture stagnates. I no more want a world run by RMS than I do a society that is beholding to Mr. Gates. A pox on both their monopolistic houses.
I believe in the competition of ideas, and strongly support and defend the concept of intellectual freedom. I've debated these issues with RMS directly, so I've been through this war before. Much as I respect his ideals, his implementation is lacking and his approach dogmatic and absolutist. Freedom involves the right to disagree with RMS and his acolytes.
The current U.S. political system does not offer significant choice, which is precisely why we need diversity. The difference between Demicans and Republocrats is so small as to be meaningless; trading Kerry for Bush is nothing more than changing dance partners while the same tune plays.
Two reasons:
Most of the compilers you list are not serious contenders. OpenWatcom is planning, but does not yet distribute, a Linux compiler. I found no mention of Linux on the lcc web site. I don't have an Alpha anymore, so the Compaq compiler is not pertinent. I don't have a copy of the Absoft compiler, even though I requested one for review. An interpreter is, as you note, not a compiler. And so on...
Comeau has contacted me by e-mail, so they'll likely be included in future updates.
Coke goes better with rum.
Well, that's my opinion. My wife insists that Diet Pepsi (ugh!) is a better mixer, but then, her sanity is in question given that she married me.
Wolfram has some interesting points, but I think his ego has gotten in the way of good science. With his "New Kind of Science", Wolfram hasn't really "invented" anything; he's mostly implying meaning where it may or may not exist.
On the other hand, I do believe that higher order derives from basic, simple, underlying processes that combine in great complexity. Turning that supposition into real science is something I hope to see happen in the next century.
Well gosh darn, your testamony makes me feel all warm and fuzzy inside. Or maybe that sensation comes from the rum and frosting...
As you noted, the problem is finding such a benchmark that doesn't bring I/O and other external latencies into play. I'm open to suggestions like yours.
Hey, I've got Ghirardelli Double Chocolate (a tasty bittersweet chip) and Jamaican Rum. Ha!
The rum probably explains the typos people found in the article. :)
Good points.
I've had problems using Intel to integrate existing libraries. As of version 8.1, the binary compatibility is supposed to be better, but I haven't had a chance to apply it to a really big project. In all fairness, GCC has had some similar problems with the C++ ABI at times.
I'm working on a Fortran 95 comparison, which will include several products.
Most people doing serious numerical work are creating proprietary or in-house applications; few people have the clusters required to run the type of code written in Fortran. So there is a commercial market for Fortran; places like National Laboratories and universities have the money to buy software.